Previous | Next --- Slide 13 of 78
Back to Lecture Thumbnails
Spantheras

Since I'm taking the deep learning class, my thoughts took that direction when these slides came up. Do you think it's possible to train a model with images like this that you generate by changing parameters and exporting the result, and for the model to then be able to predict what value on the 'diffuse color', 'metallic', 'specular light', reflectivity' etc that was used?

maq

Looks like there is work done in that! Here is an article about a CNN model that reproduce lighting or plastics, metals, wood, and other materials https://venturebeat.com/2018/08/15/researchers-develop-ai-that-can-re-create-real-world-lighting-and-reflections/